Professional Documents
Culture Documents
NursingPlus Open
journal homepage: www.elsevier.com/locate/npls
Research article
art ic l e i nf o a b s t r a c t
Article history: This paper describes the research process – from planning to presentation, with the emphasis on
Received 15 September 2015 credibility throughout the whole process – when the methodology of qualitative content analysis is
Received in revised form chosen in a qualitative study. The groundwork for the credibility initiates when the planning of the study
24 January 2016
begins. External and internal resources have to be identified, and the researcher must consider his or her
Accepted 29 January 2016
experience of the phenomenon to be studied in order to minimize any bias of his/her own influence. The
purpose of content analysis is to organize and elicit meaning from the data collected and to draw realistic
Keywords: conclusions from it. The researcher must choose whether the analysis should be of a broad surface
Content analysis structure (a manifest analysis) or of a deep structure (a latent analysis). Four distinct main stages are
Credibility
described in this paper: the decontextualisation, the recontextualisation, the categorization, and the
Qualitative design
compilation. This description of qualitative content analysis offers one approach that shows how the
Research process
general principles of the method can be used.
& 2016 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
http://dx.doi.org/10.1016/j.npls.2016.01.001
2352-9008/& 2016 The Author. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
M. Bengtsson / NursingPlus Open 2 (2016) 8–14 9
trustworthiness (Downe-Wambolt, 1992; Morse & Richards, 2002; (p. 18). Downe-Wambolt (1992) underlines that content analysis is
Patton, 2002; Silverman, 2001). This article focuses on content more than a counting process, as the goal is to link the results to
analysis and on several definitions and descriptions of content their context or to the environment in which they were produced:
analysis as a quantitative and/or qualitative method that have been “Content analysis is a research method that provides a systematic and
presented over the years. In 1952, Berelson defined content analysis objective means to make valid inferences from verbal, visual, or written
as “a research technique for the objective, systematic and quantitative data in order to describe and quantify specific phenomena” (p. 314).
description of the manifest content of communication” (p. 18). By An overview of the research process from planning to pre-
using the concepts technique and objective, Berelson underlines the sentation can be seen in Fig. 1.
process of analysis as a reliable and learnable method that pre-
cludes the personal authority of the researcher. However, Berelson's
definition does not capture the qualitative and latent perspective of The planning
the analysis. In order to make the method applicable both in a
quantitative and a qualitative approach, and without specifying the In all research, it is essential to begin by clarifying what the
depth of analysis, Krippendorff (2004) defined content analysis researcher wants to find out, from whom and how. The purpose
as “a research technique for making replicable and valid inferences may be of a descriptive or exploratory nature based on inductive or
from texts (or other meaningful matter) to the contexts of their use ” deductive reasoning. Inductive reasoning is the process of
THE IMPACT ON THE QUALITY OF THE STUDY
PLANNING
Which external and internal resources
AIM
Is there an interesting problem taped in sizeto be elucidated?
SAMPLE AND UNIT OF ANALYSIS
are available?
Who can best answer the queries set out in the aim?
METHOD OF DATA COLLECTION
How should the information from the informants be collected?
METHOD OF ANALYSIS
How should the collected data be analyzed?
PRACTICAL IMPLIKATION
Do we need someone’s approval?
DATA COLLECTION
By written or verbal questions, or by observations, and transform the
collected data into a written text
.
DATA ANALYSING
Stage 1. Decontextualisation
Identify meaning units
Create code list, repeat and start at new pages
,
Coding system Coding system
Inductive or deductive Inductive or deductive
Stage 2. Recontextualisation
Include “content”- exclude “dross”
Distance
Compare Compare
with the original data with the original data
Stage 3. Categorisation
Identify homogeneous groups
Triangulation by investigators
Stage 4. Compilation
Draw realistic conclusions
Member check, colleagues, inquiry audit
Use the words themselves Find the underlying
stay close to the text meaning of the text
developing conclusions from collected data by weaving together open-ended, written questions cannot provide the same depth that
new information into theories. The researcher analyses the text an interview can provide, as the researcher has the opportunity to
with an open mind in order to identify meaningful subjects deepen the discussion with the informants.
answering the research question. Deductive reasoning is the
opposite. Here, the researcher looks for predetermined, existing The choice of analysis method
subjects by testing hypotheses or principles (Berg, 2001; Cat-
anzaro, 1988; Polit & Beck, 2006). When a study has been initiated, In all data analysis, regardless of whether it is within a positi-
the study design has to be established. Moreover, five main issues vist or naturalistic research tradition, the purpose is to organize
must be considered in the planning process: the aim, the sample and elicit meaning from the data collected and draw realistic
and unit of analysis, the choice of data collection method, the choice conclusions (Polit & Beck, 2006). As a method, content analysis is
of analysis method and the practical implications. The researcher has unique in that it has both a quantitative (Krippendorff, 2004;
to consider these five issues repeatedly before starting the data Neuendorf, 2002) and a qualitative methodology (Berg, 2001;
collection in order to anticipate unexpected events. To ensure Burnard, 1991; Catanzaro, 1988; Downe-Wambolt, 1992), and it
credibility, there must be a chain of logic between the decisions can be used in an inductive or a deductive way. Quantitative
made and how the study will be conducted. In order to recapitu- content analysis has its origin in media research, while qualitative
late and report the development of the process, the researcher content analysis has its roots originally in social research. Despite
should write memos during this process (Morse & Richards, 2002). this, none of the forms of content analysis are linked to any par-
ticular science. Consequently, there are no specific conceptions of
The aim meaning, and the concepts used are universal. In quantitative
content analysis, facts from the text are presented in the form of
The planning discussion begins by establishing the aim, which frequency expressed as a percentage or actual numbers of key
determines the structure of the study design and sets its categories (Berelson, 1952; Krippendorff, 2004; Neuendorf, 2002).
boundaries (Downe-Wambolt, 1992). It is important to identify This method summarizes rather than reports all details concerning
an interesting and unexplored dilemma, and then present it in a a message set, and the researcher seeks to answer questions about
concise form. If the aim of the investigation is too broad, the risk how many (Krippendorff, 2004; Neuendorf, 2002). In qualitative
of touching upon too many aspects may preclude the researcher content analysis, data are presented in words and themes, which
reaching the desired depth of the studied phenomenon (Silver- makes it possible to draw some interpretation of the results. The
man, 2001). Even if the researcher can handle a large amount of choice of analysis method depends on how deep within the ana-
data in a content analysis, difficulties do arise when the purpose lysis the researcher attempts to reflect the informants' statements
is too broad. about a subject. In turn, this affects the number of informants
needed and in the way in which data are to be collected (Burnard,
The sample and units of analysis 1991; Polit & Beck, 2006). The researcher has to choose whether
the analysis is to be a manifest analysis or a latent analysis. In a
In qualitative studies, it is common that data are based on 1 to manifest analysis, the researcher describes what the informants
30 informants (Fridlund & Hildingh, 2000). However, the sample actually say, stays very close to the text, uses the words them-
size should be determined on the basis of informational needs so selves, and describes the visible and obvious in the text. In
that the research question can be answered with sufficient con- contrast, latent analysis is extended to an interpretive level in
fidence (Krippendorff, 2004; Patton, 2002). The concept unit of which the researcher seeks to find the underlying meaning of the
analysis refers to the sample, and the researcher has to determine text: what the text is talking about (Berg, 2001; Catanzaro, 1988;
whether the material is to be analyzed in its entirety or divided Downe-Wambolt, 1992).
into smaller units. The researcher has, for example, to decide
whether data collected from women and men or data collected at Practical implications
different times in a person's life should be analyzed separately or
not. In this decision, the researcher is guided by the aim to be During the planning discussion, the researcher must take
achieved, and each unit of analysis implies a different focus for the ethical aspects into consideration in order to protect the infor-
study. The key issue in making this decision is to decide what the mants. When the design is established, a presentation of these
researcher is seeking to elucidate by the study (Patton, 2002). aspects is sent to the appropriate ethics committee where
There are no established criteria when using content analysis for applicable. An information letter to persons involved in the study
the size of a unit of analysis, neither the number of informants or must be written, and all participants must be informed orally and/
objects to study, nor the number of pages based on the informants' or in writing about the study before being included. The partici-
own written text or transcribed data. pants must be guaranteed confidentiality and informed that par-
ticipation is voluntary and that they can withdraw their data from
The choice of data collection method the study at any time without it affecting their relationship with
the researchers.
Content analysis can be used on all types of written texts no
matter where the material comes from. Moreover, there are no
specific rules that must be followed (Berg, 2001; Burnard, 1991; Data collection
Catanzaro, 1988; Downe-Wambolt, 1992), for example, deep inter-
views (Wann-Hansson, Hallberg, Klevsgård & Andersson, 2005), Verbal and non-verbal interaction between informants and
focus group interviews (Golsäter, Sidenvall, Lingfors & Enskär, 2011), researchers shape the data collected, which in turn affects the
one single written question (Bengtsson, Ohlsson & Ulander, 2007), results of the study. It is important that the verbal or written
open-ended questions as in a questionnaire (Donath, Winkler, questions are appropriately formulated and adapted to the claims
Graessel, & Luttenberger, 2011), or observations of situations (East- of the referenced method so that the researcher is able to find
wood, O'Connell & Considine, 2011) as well as from pictures and some understanding of the phenomena being studied (Morse &
films (Krippendorff, 2004; Neuendorf, 2002). However, the choice of Richards, 2002). All types of questions related to the aim of the
data collection method affects the depth of the analysis. For example, study can be utilized when content analysis is used. Therefore, the
M. Bengtsson / NursingPlus Open 2 (2016) 8–14 11
Perception of gladness
ted to social isolation
2004)
2007)
transcribed into a written form, which is an interpretive process.
Transcriptions sometimes need to be very detailed to capture
Social isolation
Social isolation
speed, tone of voice, emphasis, timing and pauses, but these
Happy feelings
components can be difficult to write down and also analyze. It is
Category
Category
category
Generic-
preferable that the researcher performs the transcribing procedure
(Bailey, 2008).
Data analyzing
Glad to be alive
Sub-categories
Sub-categories
Needing toilets
Sub-headings
In a review of the literature, different opinions on the use of
concepts, procedures and interpretation in content analysis are pre-
people
sented. However, there are similarities in the way the researchers
explain the process: either they do it by using different distinguish-
ing stages, (Burnard, 1991; Downe-Wambolt, 1992), or in running
Glad to be alive
text (Berg, 2001; Catanzaro, 1988). The differences lie in the order in
Needing toilets
which the steps of analysis are taken, as well as in the way the
researcher contemplates the data and subsequently conceptualizes.
people
Code
Four main stages have been identified: the decontextualisation, the
recontextualisation, the categorisation, and the compilation (Fig. 1).
However, each stage must be performed several times to maintain
a good day. I can never make plans. I do not know how I will feel tomorrow or
When I feel bad I do nothing. I can never be really sure that tomorrow will be
When I need to run to the bath-room 5-6 times in one hour, I do not dare leave
sense of the whole, that is, to learn “what is going on?”, before it
can be broken down into smaller meaning units. A meaning unit is
the smallest unit that contains some of the insights the researcher
needs, and it is the constellation of sentences or paragraphs con-
Meaning unit
taining aspects related to each other, answering the question set out
in the aim (Catanzaro, 1988; Graneheim & Lundman, 2004). Each
identified meaning unit is labeled with a code, which should be
Table 1
analyzing process, codes facilitate the identification of concepts Burnard (1991) terms sub-headings, are the smallest units based on
around which the data can be assembled into blocks and patterns meaning units. In a manifest analysis, sometimes these are the same
(Catanzaro, 1988). The researcher should use a coding list, including as the codes of the meaning units. Sub-categories can be sorted into
explanations of the codes, to minimize a cognitive change during broader categories. The concept sub-theme can be used in a latent
the process of analysis in order to secure reliability (Catanzaro, analysis instead of the concept categories (Graneheim & Lundman,
1988; Downe-Wambolt, 1992; Morse & Richards, 2002). Codes can 2004). Identified themes and categories should be internally homo-
be generated inductively or deductively, depending on the study geneous and externally heterogeneous, which means that no data
design. If the study has a deductive reasoning design, the researcher should fall between two groups nor fit into more than one group
has to create a coding list before starting the analyzing process. (Krippendorff, 2004; Patton, 2002). Theme is an overall concept of an
Otherwise, the list can be created in the course of the process underlying meaning on an interpretative latent level, and it answers
(Catanzaro, 1988). Codes created inductively may change as the the question “How?” (Graneheim & Lundman, 2004). There are no
study progresses, as more data become available. Interpretations of informal ways to describe specific strategies when categorization is
the meaning units that seemed clear at the beginning may be performed. However, all categories must be rooted in the data from
obscured during the process. Therefore, the coding process should which they arise. Moving meaning units back and forth between
be performed repeatedly, starting on different pages of the text each categories provide progressive development of the category outcome,
time to increase the stability and reliability (Downe-Wambolt, and Post-it notes may be of help in this process. Initially, several
1992). However, it is much easier to obtain high reliability with categories are often generated, but the number is later reduced (Bur-
code lists generated deductively rather than inductively (Catanzaro, nard, 1991). How the researcher knows when the categorization is
1988). There are also computer-programs which can be of help. good enough depends on the aim of the study, and the categorization
Though their use is not imperative, they may facilitate the process. is finished when a reasonable explanation has been reached.
Although these programs do not analyze the data, they do speed up
the process by, for example, locating codes and grouping data Stage 4 The compilation
together in categories. Nevertheless, the researcher must still decide
what constitutes the themes and what conclusions can be drawn Once the categories are established, the analysis and writing up
from the results (Flick, 2002; Patton, 2002). Since computer- process begins. One difference between the various qualitative
programs are soulless software, and human creativity is of impor- analyzing methods is how the researcher relates to the analyzing
tance and needed, the human being is required. process itself and adapts to the results. In phenomenological and
hermeneutical- based studies, the researcher focuses on exploring
Stage 2 The recontextualisation how the informants make sense of experience and transform
experiences into consciousness. The researcher must then attempt
After the meaning units have been identified, the researcher to find the essence of the studied phenomenon. The researcher has
has then to check whether all aspects of the content have been the opportunity to reach a deeper understanding even if it is on a
covered in relation to the aim (Burnard, 1991). The original text is descriptive level (Patton, 2002). When performing a qualitative
re-read alongside the final list of meaning units. Colored pencils content analysis, the investigator must consider the data collected
are useful to distinguish each meaning unit in the original tran- from a neutral perspective and consider their objectivity. However,
script. After this process has been performed, unmarked text the researcher has a choice between the manifest and the latent
nearly always remains. The researcher must then consider whe- level, and the depth of the analysis will depend on how the data
ther or not the unmarked text should be included. If the unmarked are collected. In a manifest analysis, the researcher works this way
text gives some answers to the research question, it should, gradually through each identified category and through the
therefore, be included in the analysis; otherwise this “dross” can themes in a latent analysis. In a manifest analysis, the researcher
be excluded (Burnard, 1991, 1995). When the researcher is deeply often uses the informants' words, and he/she remains aware of the
involved with the data, everything seems to be of importance. need to refer back to the original text. In this way, it is possible to
Nevertheless, a process of distancing is necessary, and the stay closer to the original meanings and contexts (Burnard, 1991).
researcher must allow him or herself to let go of the unimportant In contrast, a latent analysis invites the researcher to immerse
information that does not correspond to the aim of the study. him/herself to some extent in the data in order to identify hidden
meanings in the text. For each category or theme, the researcher
Stage 3 The categorisation chooses appropriate meaning units presented in the running text
as quotations. Regardless of the form of the analysis, the
Before the researcher can begin to create categories, extended researcher can present a summary of themes, categories/sub-
meaning units must be condensed. This entails that the number of themes and sub-categories/sub-headings as a table to allow the
words is reduced without losing content of the unit (Graneheim & reader to get a quick overview of the results. In addition, it is
Lundman, 2004). The depth of the meaning units determines the appropriate to present one example of the analysis process. There
level at which the analysis can be performed. This process of is also the possibility to add information by performing some
condensation is often needed when data are based on interviews quantification in which sub-categories and categories are counted.
and when latent content analysis is to be carried out. To extract This is not normally done in other qualitative research methods.
the sense of the data, the coded material can, as a suggestion, be However, nearly everything can be counted in written messages –
divided into domains: broad groups based on different attentions such as words, characters, paragraphs and concepts – depending
of the study (Catanzaro, 1988; Patton, 2002). Graneheim and on the focus of the study. By combining the quantification with a
Lundman (2004) prefer the concept content area, since, in their qualitative approach, the magnitude of the individual phenomena
view, this elucidates a specific, explicit area. For example, the studied appears more clearly (Berg, 2001; Morgan, 1993). How-
material can be divided on the basis of the questions used when ever, the variables cannot be ranked, since not all informants have
the data were collected or on theoretical assumptions from the had the opportunity to discuss all the phenomena that the
literature. researcher finally counts.
In the categorization process, themes and categories are identified. As a final check, the researcher must consider how the new
However, in the literature there is no consensus for which headings or findings correspond to the literature and whether or not the result
concepts are to be used in a content analysis. Sub-categories, which is reasonable and logical (Burnard, 1991; Morse & Richards, 2002).
M. Bengtsson / NursingPlus Open 2 (2016) 8–14 13
To validate the outcome and to strengthen the validity of the to the objectivity or neutrality of the data (Polit & Beck, 2006). In
study, the investigator can perform a respondent validation, a general, one can say that credibility corresponds to validity,
member check, which means that the investigator goes back to the dependability to reliability and transferability to generalization. In
informants and presents the results in order to achieve agreement any case, the key is not the choice of concept but how the concepts
(Burnard, 1991; Catanzaro, 1988) However, there is a time-delay are discussed in relation to “truth” and “trustworthiness”, since in
between the data collection and the analysis. This approach, qualitative studies there is no definite “truth”. In general, the
therefore, constitutes a risk for various reasons, one of which researchers are more interested in depth understanding of a spe-
might be the possible unreliability of the informants' memory. cific issue and in showing different perspectives rather than aim-
Another risk is that informants have a tendency to deny less ing at singular truth and generalization (Patton, 2002; Rolfe,
attractive aspects of their behavior (Long & Johnson, 2000). In 2006). Consequently, qualitative studies are also sometimes
addition, as the researcher often creates a deeper holistic under- impossible and always difficult to replicate because the data arise
standing of the studied phenomenon, the informants may not from a specific context (Morse & Richards, 2002).
recognize how the data is presented (Downe-Wambolt, 1992). How the researcher handles self-reflection plays a key role in
Keeping this in mind, it is better for the researcher to obtain some all qualitative research. For example, in studies based on phe-
confirmation on the content from the informants in connection nomenological or hermeneutical methodology, the concepts
with the data collection (Catanzaro, 1988). Another way to “bracketing” and “epoche” are used to explain how the investi-
increase validity is for a colleague not involved in the study, or an gator should behave in relation to the analyzing procedure. This
inquiry auditor, to read the original text and the results and then approach affects the researcher's interpretation of the informants'
judge whether they are reasonable or not (Burnard, 1991; story and the conclusions given in the result (Morse & Richards,
Catanzaro, 1988; Downe-Wambolt, 1992). However, it is obviously 2002; Patton, 2002). In content analysis, the researcher must know
difficult for an independent person to familiarize him/herself with the context, but he/she must also be aware of this knowledge so
another's coding (Morse & Richards, 2002). that it does not affect neither process nor outcome. Certain
activities in the field remain hidden from the view of the
A discussion about concepts of trustworthiness researcher if he/she is a stranger to the context (Flick, 2002).
Bengtsson, M., Ohlsson, B., & Ulander, K. (2007). Women with irritable bowel Golsäter, M., Sidenvall, B., Lingfors, H., & Enskär, K. (2011). Adolescents' and school
syndrome and their perception of a good quality of life. Gastroenterology Nur- nurses' perceptions of using a health and lifestyle tool in health dialogues.
sing, 30(2), 74–82. Journal of Clinical Nursing, 20, 2573–2583.
Berelson, B. L. (1952). Content analysis in communications research. New York: Free Graneheim, U. H., & Lundman, B. (2004). Qualitative content analysis in nursing
press. research: concepts, procedures and measure to achieve trustworthiness. Nurse
Berg, B. L. (2001). Qualitative research methods for the social sciences. Boston: Allyn Education Today, 24, 105–112.
and Bacon. Krippendorff, K. (2004). Content analysis: an introduction to its methodology.
Burnard, P. (1991). A method of analysing interview transcripts in qualitative Thousand Oaks, California: Sage Publications Inc.
research. Nurse Education Today, 11, 461–466. Lincoln, YS. & Guba, EG. (1985). Naturalistic Inquiry. Sage Publications, Newbury
Burnard, P. (1995). Interpreting text: an alternative to some current forms of textual Park, CA.
analysis in qualitative research. Social Sciences in Health, 1, 236–245. Long, T., & Johnson, M. (2000). Rigour, reliability and validity in qualitative research.
Catanzaro, M. (1988). Using qualitative analytical techniques In: N. F. Woods, & Clinical Effectiveness in Nursing, 4, 30–37.
M. Catanzaro (Eds.), Nursing: research theory and practice (pp. 437–456). St. Morgan, D. (1993). Qualitative content analysis: a guide to paths not taken. Quali-
Louis: The CV Mosby Company. tative Health Research, 3, 112–121.
Donath, C., Winkler, A., Graessel, E., & Luttenberger, K. (2011). Day care for Morse, J. M., & Richards, L. (2002). Read me first for a user's guide to qualitative
dementia patients from a family caregiver's point of view: a questionnaire methods. Thousand Oaks, California: Sage Publications Inc.
study on expected quality and predictors of utilisation – Part II. BMC Health Neuendorf, K. (2002). The content analysis guidebook. Thousand Oaks, California:
Services Research, 2011(11), 1–7. Sage Publications Inc.
Downe-Wambolt, B. (1992). Content analysis: method, applications and issues. Patton, M. Q. (2002). Qualitative, research & evaluation methods. Thousand Oaks,
Health Care for Women International, 13, 313–321. California: Sage publications Inc.
Eastwood, G. M., O'Connell, B., & Considine, J. (2011). Low-flow oxygen therapy in Polit, D. F., & Beck, C. T. (2006). Essentials of nursing research methods, appraisal, and
intensive care: an observational study. Australian Critical Care, 24, 269–278. utilization. Philadelphia, PA: Lippincott Williams & Wilkins.
Elo, S., & Kyngäs, H. (2007). The qualitative content analysis process. J Adv Nurs., 62 Rolfe, G. (2006). Validity, trustworthiness and rigour: quality and the idea of qua-
(1), 107–115. litative research. Journal of Advanced Nursing, 53, 304–310.
Elo, S., Kääriäinen, M., Kanste, O., Pölkki, T., Utriainen, K., & Kyngäs, H. (2014). Silverman, D. (2001). Interpreting qualitative data methods for analyzing talk, text and
Qualitative content analysis: a focus on trustworthiness. SAGE Open, 4. interaction. Thousand Oaks, California: Sage publications Inc.
Flick, U. (2002). An introduction to qualitative research. London: Sage Publications Wann-Hansson, C., Hallberg, I. R., Klevsgård, R., & Andersson, E. (2005). Patients
Inc. ´experiences of living with peripheral arterial disease awaiting intervention: a
Fridlund, B., & Hildingh, C. (2000). Health and qualitative analysis methods In: qualitative study. International Journal of Nursing Studies, 42, 851–862.
B. Fridlund, & C. Hildingh (Eds.), Qualitative research, methods in the service of
health (pp. 13–25). Lund: Studentlitteratur.